Language-Specific Layer Matters: Efficient Multilingual Enhancement for Large Vision-Language Models
Comments Accepted by EMNLP 2025 findings
期刊&会议
EMNLP
Conference on Empirical Methods in Natural Language Processing · 会议 · Natural Language Processing
Can VLMs Recall Factual Associations From Visual References?
机构 * University of Southern California（南加州大学） ； Information Sciences Institute, University of Southern California（信息科学研究所，南加州大学）
Comments To appear at EMNLP 2025 (Findings)
SDGO: Self-Discrimination-Guided Optimization for Consistent Safety in Large Language Models
Comments Accepted by EMNLP 2025 (Main Conference), 15 pages, 4 figures, 6 tables
Dense Retrievers Can Fail on Simple Queries: Revealing The Granularity Dilemma of Embeddings
机构 * Pattern Recognition Center, WeChat AI（微信AI模式识别中心）
Comments Accepted to EMNLP 2025 Findings
Jigsaw-Puzzles: From Seeing to Understanding to Reasoning in Vision-Language Models
机构 * Hangzhou Institute for Advanced Study, UCAS（杭州高等研究 institute，UCAS） ； Zhejiang Lab（浙江实验室） ； Zhejiang University（浙江大学）
Comments Accepted by EMNLP 2025 Main conference
Debate-to-Detect: Reformulating Misinformation Detection as a Real-World Debate with Large Language Models
机构 * School of Advanced Interdisciplinary Sciences, University of Chinese Academy of Sciences（中国科学院大学先进交叉学科学院） ； State Key Laboratory of Mathematical Sciences, Academy of Mathematics and Systems Science, Chinese Academy of Sciences（中国科学院数学与系统科学研究院）
Comments This paper has been accepted to EMNLP 2025 (Main Conference)
SolEval: Benchmarking Large Language Models for Repository-level Solidity Code Generation
Comments Accepted By EMNLP'25-Main
Aligning NLP Models with Target Population Perspectives using PAIR: Population-Aligned Instance Replication
机构 * University of Maryland, College Park（马里兰大学 College Park 分校） ； LMU Munich & Munich Center for Machine Learning（慕尼黑大学及慕尼黑机器学习中心） ； RTI International（RTI国际）
Comments EMNLP 2025 NLPerspectives Workshop
TL-Training: A Task-Feature-Based Framework for Training Large Language Models in Tool Use
机构 * Fudan University（复旦大学） ； Lenovo Research（联想研究） ； ByteDance Seed（字节跳动种子） ； Shanghai Key Lab of Intelligent Information Processing（上海智能信息处理重点实验室） ； Shanghai AI Laboratory（上海人工智能实验室）
Comments Accepted by EMNLP 2025
Label Set Optimization via Activation Distribution Kurtosis for Zero-shot Classification with Generative Models
机构 * Department of Computer Science, University of Sheffield, UK（计算机科学系，谢菲尔德大学）
Comments Accepted by EMNLP 2025
fLSA: Learning Semantic Structures in Document Collections Using Foundation Models
机构 * Microsoft Research Redmond（微软研究院雷德蒙德）
Comments EMNLP 2025 Camera Ready
Recognizing Limits: Investigating Infeasibility in Large Language Models
机构 * University of California Irvine（加州大学尔滨分校）
Comments EMNLP 2025 Findings
Learning from Few Samples: A Novel Approach for High-Quality Malcode Generation
机构 * Huazhong University of Science and Technology（华中科技大学） ； Peking University（北京大学）
Comments 18pages,5 figures,emnlp
Understanding Subword Compositionality of Large Language Models
机构 * University of Copenhagen（哥本哈根大学） ； ETH Zurich（苏黎世联邦理工学院）
Comments EMNLP 2025 Main
Debiasing Multilingual LLMs in Cross-lingual Latent Space
机构 * University of Copenhagen（哥本哈根大学） ； ETH Zurich（苏黎世联邦理工学院）
Comments EMNLP 2025 Main
Riemannian Optimization for LoRA on the Stiefel Manifold
机构 * Opt-AI Inc.（Opt-AI公司）
Comments EMNLP 2025 Findings
Speech Discrete Tokens or Continuous Features? A Comparative Analysis for Spoken Language Understanding in SpeechLLMs
机构 * The Chinese University of Hong Kong（香港中文大学）
Comments Accepted to EMNLP 2025 Main Conference
Beyond Demographics: Enhancing Cultural Value Survey Simulation with Multi-Stage Personality-Driven Cognitive Reasoning
机构 * School of Computer Science and Technology, Wuhan University of Science and Technology（武汉科技大学计算机科学与技术学院） ； Hubei Province Key Laboratory of Intelligent Information Processing and Real-time Industrial System, Wuhan University of Science and Technology（湖北省智能信息处理与实时工业系统重点实验室） ； Innovation, Policy and Entrepreneurship Thrust, The Hong Kong University of Science and Technology (Guangzhou)（香港科技大学（广州）创新、政策与创业方向） ； Tübingen AI Center, University of Tübingen（图宾根大学图宾根人工智能中心） ； Department of Computer Science, University of Copenhagen（哥本哈根大学计算机科学系）
Comments 23 pages, 6 figures, accepted to EMNLP 2025 main
Weights-Rotated Preference Optimization for Large Language Models
机构 * Institute of Information Engineering, Chinese Academy of Sciences（中国科学院信息工程研究所） ； School of Cyber Security, University of Chinese Academy of Sciences（中国科学院大学网络安全学院） ； Baidu Inc.（百度公司）
Comments EMNLP 2025
The Arabic Generality Score: Another Dimension of Modeling Arabic Dialectness
机构 * MBZUAI ； New York University Abu Dhabi（纽约大学阿布扎比分校）
Comments Accepted to EMNLP 2025 Main Conference
Quantifying Language Disparities in Multilingual Large Language Models
机构 * Language Technology Lab, University of Cambridge, UK（剑桥大学语言技术实验室）
Comments Accepted at EMNLP 2025
SPORTSQL: An Interactive System for Real-Time Sports Reasoning and Visualization
机构 * Arizona State University（亚利桑那州立大学）
Comments Under Review at EMNLP
The Power of Framing: How News Headlines Guide Search Behavior
机构 * Department of Computer Science and Engineering（计算机科学与工程系） ； ND-IBM Technology Ethics Lab（ND-IBM技术伦理实验室） ； University of Notre Dame（北达科他大学）
Comments Accepted to EMNLP
DeAR: Dual-Stage Document Reranking with Reasoning Agents via LLM Distillation
机构 * University of Innsbruck（因斯布鲁克大学）
Comments Accept at EMNLP Findings 2025
Mixture of Length and Pruning Experts for Knowledge Graphs Reasoning
机构 * The Hong Kong University of Science and Technology (Guangzhou)（香港科学与技术大学（广州））
Journal ref Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing (EMNLP), Suzhou, China
Self-Correcting Code Generation Using Small Language Models
机构 * Graduate School of Artificial Intelligence, POSTECH（人工智能研究生院，POSTECH） ； Department of Computer Science and Engineering, POSTECH（计算机科学与工程系，POSTECH）
Comments Accepted at EMNLP 2025 (Findings, long paper)
DecisionFlow: Advancing Large Language Model as Principled Decision Maker
机构 * University of Illinois at Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校）
Comments EMNLP 2025 Findings; 25 pages, 15 figures
Exploring the Vulnerability of the Content Moderation Guardrail in Large Language Models via Intent Manipulation
机构 * Boise State University（博伊州立大学） ； University of Illinois Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校） ； University of Pittsburgh（匹兹堡大学） ； New York University（纽约大学） ； Florida International University（佛罗里达国际大学）
Comments Accepted for EMNLP'25 Findings. TL;DR: We propose a new two-stage intent-based prompt-refinement framework, IntentPrompt, that aims to explore the vulnerability of LLMs' content moderation guardrails by refining prompts into benign-looking declarative forms via intent manipulation for red-teaming purposes
TombRaider: Entering the Vault of History to Jailbreak Large Language Models
机构 * UNSW（新南威尔士大学） ； NTU（国立大学）
Comments Main Conference of EMNLP
PRISM: Efficient Long-Range Reasoning With Short-Context LLMs
Comments Published as a conference paper at EMNLP 2025. 28 pages, 7 figures, 5 tables
